Registries, IVO ids, and Data Set Identifiers (fwd)

[Tony Linde]

Sounds good to me. 

The only mismatch with the current WD is that the resource key should not
have '.'s but use only '/'s as the separator. But given that we can
distinguish the authority id from the resource key using the *first* '/' in
the string, I have no problem with allowing '.' in the resource key - makes
it look less like a URL as well which is all to the good.

Cheers,
Tony. 

> -----Original Message-----
> From: owner-registry at eso.org [mailto:owner-registry at eso.org] 
> On Behalf Of Robert Hanisch
> Sent: 25 September 2003 01:58
> To: Arnold Rots; registry at ivoa.net
> Subject: Re: Registries, IVO ids, and Data Set Identifiers (fwd)
> 
> 
> Does this idea make sense...?
> 
> The suite of telescope labels could function, and perhaps 
> might more properly function, as resource keys under a common 
> authority, such as
> 
>     ivo://ads.harvard.edu/sa.cxo#2000
> 
> This allows the proposed dataset IDs to map virtually as-is 
> onto the VO registry Identifier, prefixed only by an 
> authority ID which asserts, properly, that an agreement with 
> the journals and data centers defines the range of resource 
> keys.  It seems to me that this solves the persistence 
> problem, as the registries (or some other service) can be 
> used to resolve the ID into its associated URL, or service, 
> or whatever.  Here I have assumed that the ADS would act as 
> the authority, though other possibilities come to mind, such 
> as having IVOA itself assume this role for such a broad 
> matter.  In that case we might have
> 
>     ivo://ivo.net/sa.cxo#2000
> 
> However, given the well-established role of the ADS I think 
> the first form is preferable.
> 
> The drawback, perhaps, is that there are now at least two 
> identifiers for a particular data set, such as
> 
>     ivo://sao.harvard.edu/cxo#2000
> and
>     ivo://ads.harvard.edu/sa.cxo#2000
> 
> since both data centers and ADS/astronomy journals will wish 
> to publish their collections in the registry.  We will have 
> different identifiers already, though, for mirrors, so this 
> in itself is not a fatal problem.  The RM schema includes 
> elements Relationship and RelationshipID that can be used to 
> explicitly capture the "mirror-of" status, as in
> 
>     Identifier = ivo://sao.harvard.edu/cxo#2000
> 
> and another,
> 
>     Identifier = ivo://ads.harvard.edu/sa.cxo#2000
>     Relationship = mirror-of
>     RelationshipID = ivo://sao.harvard.edu/cxo#2000
> 
> I don't think it is absolutely necessary to indicate the 
> relationship shown above, but the mechanism is available to us.
> 
> Is this is a way to reconcile the needs of the generic VO 
> registry with the needs of the journal article/dataset links?
> 
> Bob
>